Code and data associated with the paper "Building Bridges: A Dataset for Evaluating Gender-Fair Machine Translation into German."
- Create a new python environment (python >= 3.9 is recommended)
- Install the minimum requirements from
requirements.txt
Data is released under the CC BY 4.0 license.
/data
contains the seed nouns for our Gender-Fair Dictionary and the Multi-Sentence Multi-Domain mentions in Wikipedia and Europarl./results
contains results for our evaluation of machine-translated passages.
We release several bash and python scripts to replicate our experiments. Head to /bash
or /src
to find most relevant details.
Script names and parameters are mostly self-explanatory, but do not esistate to open an issue to ask for more details (this is the preferred way over email).